Journal of the Lepidopterists Society 53(2),1999, 65-71

PRELIMINARY ESTIMATES OF DIVERSITY FROM SPECIFIC SITES IN THE NEOTROPICS USING COMPLEMENTARITY AND SPECIES RICHNESS ESTIMATORS

MICHAE L G. POGUE Systematic Entomology Laboratory, P. S. I. , Agricultural Research Service, U.S. D epartment of Agriculture, % D ep artment of Entomology, National Museum of Natura! HistOIY, Smithsonian Institution, Washington, D.C. 20560-0168, USA

ABSTRACT. Lepidoptera were collected and species richness and compl ementarity or uniqueness were compared between two rainfor­ est sites: Pakitza, Peru and Beni , Bolivia. The total number of species collected from hoth sites was 1,879 of which 60 were shared resulting in a complementarity of96.8%. Non-parametric equations and species accumulation curves of Hemiceras Guenee (Lepidoptera: ) were used to compare species richness between three rainforest sites, Pakitza and Tambopata, Peru and Reserva Ethnica Waorani, Onkone Gare, Ecuador. Cluster analysis, using complementarity values for selected sites was used to determine altitudinal relationships between sites in Costa Rica; relationships between forest types in Brazil; and faunal differences among sites in western Amazonia using Hemiceras. Additional key words: Biodiversity, Notodontidae, Hemiceras, Chao, jackk'l1ife .

Biodiversity as defined by E. O. Wilson (Reaka­ taxa around the globe. After a site has been sampled, Kudla et al. 1997) is "everything". Biodiversity encom­ species richness estimates are used to predict how passes the genes within a Single local population or many species were missed during the sampling pro­ species, the species within a local community, and cess, thus arriving at an estimated number of species communities comprising the diverse ecosytems of the based on the actual number observed plus the number world. Life on earth is supported by the interactions missed. By using species lists, either generated by and products produced by all other life on earth. With­ sampling at a site or from museum collections, species out biodiversity there would be no life on earth as we composition among sites can be compared. These know it. Therefore, it is essential that biologists begin comparisons can then be used in setting policy and to document and record biodiversity, whether it be making informed conservation and management deci­ how many species of are in your backyard to sions. how many species of trees in a forest to how many for­ The goals of this paper are 1) to emphaSize the im­ est types in the world. portance of adequate sampling, 2) to assess whether Anyone with an interest in natural history can begin inadequate sampling can still be useful in predicting to study their local biodiversity and to document it. species richness and complementarity between study One basic element of biodiversity is to know how sites, and 3) to prOvide a method of using complemen­ many species are present at a particular site. This pa­ tarity to compare faunal relationships between sites. per will outline how anyone can begin to document lo­ To accomplish these goals, I compared overall species cal biodiversity by gathering data on species richness richness and complementarity of Lepidoptera from or the number of species present at a site at a particu­ two rainforest sites, one in Peru and the other in Bo­ lar point in time. KnOwing what species are present at livia. In addition, I estimated species richness from a site is essential because it is the first step in under­ species accumulation curves and non-parametric esti­ standing the interactions between the species docu­ mators to compare the Hemiceras Guenee (Notodon­ mented and their interactions within the local commu­ tidae) fauna between 2 sites in Peru and 1 in Ecuador. nity and ecosystem. Finally, museum specimens of Hemiceras were used Presently, no one has much of an idea exactly how for faunal comparisons among sites in Costa Rica, many species are present today on earth. Estimates of Brazil, and western Amazonia, using complementarity the number of species worldwide range from 3 to 100 and cluster analysis. million (Erwin 1982, 1983, Stork 1988, Hodkinson & Casson 1991, May 1992, Raven and Wilson 1992). The MATERIALS AND M E THODS study of species richness and complementarity, or how Lepidoptera complementarity in SW Amazo­ different species composition is between sites, is es­ nia. Lepidoptera were collected at two rainforest sites: sential to assessing global biodiversity patterns. Beni, Bolivia and Pakitza, Peru. The Beni study site at To address the question of world diversity 14°49' S, 66°28'W is 40 km E of San Borja, at 250 me­ one must get accurate estimates of Site-specific species ters elevation. Pakitza is located on the Rio Manu at richness for a variety of taxa (Colwell & Coddington 11 °56'4TS, 71°1 TOO'W within the large drainage 1994), and then to compare these species lists to mea­ basin of the Rio Madre de Dios in southeastern Peru, sure relative levels of overlap and richness of these at 356 m elevation, approximately 550 km NW of Beni. 66 JOURNAL OF THE LEPIDOPTERISTS' SOCIETY

40

35

30

25

10 -+- Pakitza ;J ____ Tambopata

~ Onkone Gare 5

Trap Nights

FIG. 1. Species accumulation curves of Hemiceras at Pakitza and Tambopata, Peru, and Onkone Care, Ecuador.

Samples were collected between August 26-Septem­ The complementarity between the two sites is the ber 15, 1987 (Beni) and from September 27-0ctober proportion of the unique species to the pooled rich­ 5, 1987 (Pakitza). Adult were collected by UV ness, or light traps, spread, identified to family, sorted to mor­ pho-species, and counted. Voucher specimens have (3) been deposited in the National Museum of Natural History, Smithsonian Institution, Washington, D .C. Collecting effort was defined as the number of t~\lP ' Hemiceras species richness in western Amazo­ nights: 12 and 8 trap nights in Pakitza and Beni, re­ nia. The Hemiceras (Lepidoptera: Notodonti­ spectively. Other studies have included the number of dae), representing 245 species, was used as an indica­ person-hours spent collecting (Coddington et al. 1991, tor group for estimating species richness at three Robbins et al. 1996), number of collecting days (Lou­ rainforest sites: Pakitza and Rio Tambopata Research ton et al. 1996), or trap nights (approximately 12 hours Station in southeastern Peru, and Onkone Care, in in length). Ecuador. The Lepidoptera faunas of Pakitza and Beni were Rio Tambopata Research Station, at 14°14'S, compared using complementarity (Colwell and Cod­ 69°11'W, is located on the RIO Tambopata, 30 air km dington 1994). In comparing two sites, j and k, the SE of Puerto Maldonado, Madre de Dios, Peru, at first site has a species richness of Sj and the second 290 m elevation. Onkone Care, at 00038'S, 76°36W, is site has Sk If the number of species in common be­ a research station within the Reserva Ethnica Waorani, tween both sites is Vjk, then the total species richness Ecuador, at 220 m elevation. Tambopata had a total of for both sites is 29 trap nights from November 2-25, 1979 and Sep­ tember 16-21, 1990. The 11 trap nights at Onkone (1) Care were January 10, 12,13-18,25; June 20; and July 16,1994. and the number of species unique to both sites (Uik ) is Species accumulation curves (Fig. 1) were used to plot the cumulative number of new species collected (2) over unit effort (number of trap nights) at each of the VOLUME 53, N UMBER 2 67

FTC. 2. Localiti es of Costa Rican sites used in cluster analysis of dissimilarity matrix in Table 3. FIG. 3. Localities of South American sites used in cluster analyses of dissimilarity matrices in Tables 4 and 5.

three sites. To extrapolate total species richness from jackknife estimator of species richness is based on the the species accumulation curves at each site, four dif­ number of species that occur in only one sample L, ferent nonparametric equations were compared: 1) Chao 1, 2) Chao 2, 3) first-order jackknife, and 4) sec­ 1 jackknife = Sobs + L(n - lin), (6) ond-order jackknife. Chao (1984) developed a simple estimator of the where n is the number of samples. true number of species at a given site based on the The second-order jackknife (Burnham & Overton number of rare species in the pooled sample j. This is 1978,1979) is like the Chao 2 estimator where L is the considered an abundance-based estimator because it is number of species that occur in only one sample and based on the number of species that are only repre­ M is the number of species that occur in exactly two sented by only 1 or 2 individuals to estimate overall samples, species richness. Colwell and Coddington (1994) called this Chao 1, 2 jackknife = Sobs + 1/2 [L(2n - 3)/n - M(n - 2)2/n(n - 1)], Chao 1 = Sobs + a2/2b, (4) (7) where Sobs is the observed number of species in a sam­ where n is the number of samples. ple, a is the number of species that are only repre­ Hemiceras faunal comparisons between three sented by one specimen in the pooled sample (Single­ tropical regions. SpeCies lists of Hemiceras were com­ tons), and b is the number of species represented by piled from specimens in the National Museum of Natu­ two specimens in the pooled sample (doubletons). ral History, Smithsonian Institution, Washington, D.C., This estimator works well when the samples contain a and were used to examine faunal relationships among large number of rare species (Chao 1984), which fre­ sites in Costa Rica, Brazil, and western Amazonia. In quently occurs when sampling diverse groups such as Costa Rica, six sites were chosen to see how altitude af­ insects. fected species composition. The sites chosen were Juan A related estimator is Chao 2 (Colwell & Codding­ Vinas (1500 m), Tuis (732 m), Turrialba (634 m), ton 1994), which is based on the incidence ofrare spe­ Cuapiles (259 m), La Selva (40 m) and Sixaola River (0 cies among samples, m) (Fig. 2). Six sites in Brazil were chosen to examine the effect of forest type: the lowland Amazonia forest in­ Chao 2 = Sobs + U/2M, (5) cluded the sites of Sao Paulo de Oliven<;a (Amazonas) and Porto Velho (Rondonia), and the Atlantic coast for­ where L is the number of species that occur in only est sites were Baixo Cuandu (Espirito Santo), Campo one sample, and M is the number of species that occur Bello (Rio de Janeiro), Hansa Humboldt (Santa Cata­ in exactly two samples. rina), and Santa Catherines (Santa Catarina) (Fig. 3, Jackknife estimators (Burnham & Overton 1978, sites 1-6). Six sites were chosen to determine faunal re­ 1979) also use the distribution of species among sam­ lationships in western Amazonia. They included ples (Colwell & Coddington 1994). The first-order Neblina, Venezuela; Onkone Care, Ecuador; Sao Paulo 68 JOURNAL OF THE LEPIDOPTERISTS' SOCIETY

TABLE 1. Comparison of % complementarity and number of spe­ TABLE 2. Estimated total species richness of Hemiceras species cies of Lepidoptera between Beni, Bolivia and Pakitza, Peru. for 4 nonparametric estimators.

Family % Complementanty N umber shared species Species ri chness Pakitza Tam bopata Onkone Care estimator Peru Peru Ecuador Microlepidoptera Cosmopterygidae 100.0 50 (0) Observed 36 24 22 Tineidae 100.0 113 (0) Chao 1 43 29 40 Gelechiidae 98.8 167 (2) Chao 2 39 25 24 Oecophoridae 97.8 324 (7) 1 Jackknife 41 24 24 Pyraiidae/Crambidae 96.8 281 (9) 2 Jackknife 39 25 23 Macrolepidoptera Noctuidae 98.6 296 (4) Notodontidae 95.2 63 (3) Geometridae 93.3 213 (14) closer to the observed values than did Chao 1, on Arctiidae 91.5 189 (16) abundance-based estimates (Table 2). Hemiceras faunal comparisons between three tropical regions. Costa Rica: The high altitude sites de Olivenc;a and Porto velho in Brazil; and Pakitza and ofJuan Vinas and Tuis (1500 m and 732 m) were clus­ Tambopata in Peru (Fig. 3, sites 1-2, 7- 10). tered at a dissimilarity value of 0.389, and the low alti­ Complementarity was calculated between sites. tude sites (0 m and 40 m) of Sixaola River and La Selva These values were used to produce dissimilarity matri­ were clustered at a dissimilarity value of 0.571. ces which were converted, for easier interpretation, into Cuapiles (259 m) was clustered with the low sites at a dendrograms using cluster analysis (SYSTAT 1992). dissimilarity value of 0.667. The high altitude and low altitude sites were clustered at a dissimilarity value of RESULTS 0.718. Turrialba (634 m) showed the greatest faunal Lepidoptera complementarity in SW Amazonia. diSSimilarity between all other sites at 0.758 (Table 3; Some 38 families of Lepidoptera were collected from Fig. 4). both sites, although only nine families had numbers of Brazil: The two Amazonian sites of Sao Paulo de species sufficient to illustrate trends in complementarity. Olivenc;a and Porto Velho were clustered at a dissimi­ Five of these families were Microlepidoptera and the larity value of 0.883. Within the Atlantic Coast sites remaining four were Macrolepidoptera. A total of 1748 Hansa Humboldt and Santa Catherines were least dis­ specimens representing 933 species were collected at similar (0.600), Campo Bello was most similar to Beni and 1731 specimens representing 1006 species Hansa Humboldt + Santa Catherines (0.724), and from were collected at Pakitza. The pooled species Baixo Cuandu was most dissimilar to the previous At­ richness for both sites (Sjk) was 1879, the total number lantic Coast Forest sites (0.833). Dissimilarity between of unique species (Ujk ) was 1819, resulting in a com­ the Amazonia and Atlantic Coast Forests was 0.951 plementarity of96.8%. The Microlepidoptera families (Table 4; Fig. 5). had higher complementarity values (100-96.8%), than Western Amazonia: Pakitza and Tambopata in the Macrolepidoptera families (95.2-91.5%), with the southeastern Peru were clustered at a dissimilarity exception of the Noctuidae (98.6%) (Table 1). value of 0.766. Neblina, Venezuela and Onkone Care, Hemiceras species richness in western Amazo­ Ecuador were clustered at a dissimilarity value of nia. Comparing the four nonparametric equations for 0.793. These four sites were clustered at a dissimilarity estimating species richness of Hemiceras at Pakitza value of 0.806. The fauna of Sao Paulo de Olivenc;a had and Tambopata, Peru, and at Onkone Care, Ecuador a dissimilarity value of 0.815 compared to the previous resulted in Chao 1 estimating the highest number of four sites and Porto Velho's fauna was the most dissim­ species. The incidence-based estimators (Chao 2 and ilar (0.833) (Table 5; Fig. 6). the Jackknifes) consistently estimated total richness TABLE 4. Dissimilarity matrix of comparison of Brazilian Amazo­ TABLE 3. Dissimilarity matrix of Costa Rican altitudinal sites. nia and Atlantic Coast Forest.

Juan Vinas Sao Paulo de Porto Baixo Campo St. (1500m) Tuis Turrialba Guapil es La Selva OIIVenc;:a Velho Guandu Bello Catherine~ Tuis (732 m) 0.389 Porto Velho 0.883 Turrialba (634 m) 0.771 0.758 Baixo Guandu 0.982 0.951 Guapiles (259 m) 0.897 0.829 0.960 Campo Bello 0.927 0.930 0.833 La Selva (40 m) 0.718 0.730 0.897 0.769 St. CatheIines 0.945 1.000 0.867 0.724 Sixaola River (0 m) 0.794 0.781 0.909 0.667 0.571 Hansa Humboldt 0.958 1.000 0.870 0.792 0.600 VOLUME 53, NUMBER 2 69

Turrialba 634 m (1) ------,

Tuis 732 m (2) ------,

Juan Vinas 1500 m (3) ------'

La Selva 40 m (4) ------.,

Sixola River 0 m (5) ______---J 4 Guapiles 259 m (6) ------' 0.667 0.758 I I I 0389 0.571 0.718

FIG. 4. Dendrogram for cluster analysis of dissimilarity matrix (Table 3) of Costa Rican sites. Numbers refer to localities on Fig. 2. Altitudes are indicated for each site. The scale indicates dissimilarity value.

DISCUSSION suitable richness estimator) should reach an asymptote or remain constant over time with additional sampling. Lepidoptera complementarity in SW Amazo­ Of the three sites, the species accumulation curve nia. The lower complementarity found in most reached an asymptote, only at Pakitza (Fig. 1), sug­ Macrolepidoptera, compared to the Microlepidoptera, gesting that the species estimate there should be the may be due to sampling bias; because the collecting most accurate. Although Tambopata is known for the was done by UV light, larger moths may be coming high species richness of various insect groups (Fisher from a larger collecting universe than the smaller 1985, Paulson 1985, Pearson 1985, Wilkerson & moths, because the the larger species are better able Fairchild 1985, Robbins et a1. 1996). The Hemiceras to disperse than the smaller ones. The relatively high fauna there is poor, considering that there are 245 complementarity of the Noctuidae is curious given species in the neotropics. Although samples taken at that they are generally strong Biers, medium to large Tambopata and Onkone Care were insufficient to ac­ moths, and many species are known for their migra­ curately estimate richness, as shown by the non­ tion and wide ranging dispersal abilities. The answer asymptotic species accumulation curves (Fig. 1), it ap­ may lie in their diversity; because these moths are the pears from all the nonparametric estimates, and the most speciose lepidopteran family, the sampling time steeper accumulation curve, that Pakitza is even richer possible in this preliminary study may be inadequate in Hemiceras than is Tambopata. That contrasts with a to accurately assess the ranges of many noctuid spe­ study of the faunal relationships between the Ci­ cies, resulting in a higher complementarity value. cadoidea (Homoptera) (Pogue 1996) and Odonata Hemiceras species richness in western Amazo­ (Louton et a1. 1996) at Pakitza and Tambopata, in nia. To accurately estimate the total number of species which the species richness was greater at Tambopata. of a target taxon (such as Hemiceras) at a particular Thus, either the preliminary estimate for Hemiceras at site the species accumulation curve (or the curve of a Tambopata is inaccurate, or this study demonstrates

Sao Paulo de Oliven~a (1) ______--,

Porto Velho (2) ------'

Baixo Guandu (3) ------,

Campo Bello (4) ------,

S1. Catherines (5)

Hansa Humboldt (6) ------' 5 0.883 J J 0.600 0.724 0.833 0.951

FIG. 5. Dendrogram for cluster analysis of dissimilarity matrix (Table 4) of Brazilian Amazonian and Atlantic Coast Forcst sites. Numbers refer to localities on Fig. 3. The scale indicates dissimilarity value. 70 J OURNAL OF THE LEPIDOPTERlSTS' SOCIETY

Porto Velho (2) ------,

Sao Paulo de Oliven(:a (1) ------,

Pakilza (7) ------,

Tambopata (8) ______----l

Neblina (9) ------, 6 Onkone Gare (10) 0.806 0.833 I I I I 0.766 0.7930.815

FIG. 6. Dendrogram for cluster analysis of dissi milarity matrix (Table 5) of western Amazonian sites. N umbers refer to localities on Fig. 3. The scale indicates di ssimilarity value.

that different taxa, with divergent biologies, have dif­ taxon can come from collecting, or from using mu­ ferent centers of diversity. seum collections to obtain faunistic data from speCific At the Onkone Care site in Ecuador it is not clear sites. An advantage of using museum collections are whether the species accumulation curve is approach­ the data available from sites that are no longer pristine, ing an asymptote (Fig. 1) and there are more rare spe­ such as those in Amazonia that were collected more cies (12 singletons) than at the other sites, so the pre­ than 50 to 100 years ago. Today, with the destruction diction of 40 species by Chao 1 (Table 3) may be of the rain forest, these sights will no longer have the accurate. Based on the species richness data of same biota. The target taxon has to be common Hemiceras, the following are recommended for follow­ throughout the study area so it can be easily sampled up studies: 1) collecting effort must be adequate for and there should not be a dominance of rare species. the species accumulation curve, or the estimate curve, Hemiceras faunal comparisons between three to reach an asymptote, 2) if there is a preponderance tropical regions. The six sites in Costa Rica show a of rare species, (singletons) Chao 1 should give the broad altitudinal range from 0- 1500 meters. Altitude highest, and perhaps best estimate, and 3) if the num­ seems to influence species composition among sites in ber of rare species is low, Chao 2 or the second-order Costa Rica more than distance between sites. Juan jackknife may give a better estimate. It is also impor­ Vinas (1500 m) and Tuis (732 m) are the highest sites tant to choose your target taxon carefully. and are clustered at a dissimilarity value of 0.389, the To assess taxon richness, target taxa can be any cat­ lowest of any pair. If distance was the Significant limit­ egory, an order, family, subfamily, tribe, or genus. The ing factor of faunal composition, one would expect that taxon should be chosen with care. For example, one Tuis and Turrialba would be clustered. The same is that is too speciose requires too much time to process true for the lower altitudinal sites, with La Selva (40 and extrapolate the needed data, one with too few spe­ m) and Sixaola River (0 m) having the lower dissimi­ cies could result in insufficient data. I have found in larity value (0.571 ), even though La Selva is closer to the Neotropics that a target taxon of 200-400 species Cuapiles than the Sixaola River site (Table 3) (Fig. 4). seems to be large enough so that the species accumu­ Analysis of the six sites within Brazil were used to lation curve reached an asymptote after 20-30 sam­ show if there was a faunal difference between Amazo­ ples (trap nights, in this case). The data for the target nia and the Atlantic Coast forest. The Atlantic Coast sites were dustered (Baixo Cuandu to Ransa Hum­ boldt) and were quite distinct from the Amazonian

TABLE 5. Dissimilarity mattix of Amazonian sites. fauna (Sao Paulo de Olivenr,;a and Porto Velho) which were also clustered. Within the Atlantic Coast Forest Onkonf' Sao Pau lo de Porto distance seems to be influencing the faunal relation­ Neblma C

two sites, Cluster analysis indicates that there is a fau­ CODDINGTO N, J. A. , C. E. GRISWOLD, D. SILVA DAVILA, E. PE­ nal difference between Amazonia and the Atlantic NARAN DA & S. F. LARCHER. 1991. Designing and testing sam­ pling protocols to estimate biodiversity in tropical ecosystems, Coast Forest (Table 4) (Fig. ,5). pp. 44-60. In E. C. Dudley (ed.), The unity of evolutionmy bi­ Amazonia is often treated as one large biogeo­ ology: Proceedings of the Fourth International Congress of Sys­ graphical area, but just how different are sites within tematic and Evolutionary Biology. Dioscorides Press, Portland, Oregon. 1048 pp Amazonia? Among the six Amazonian localities dissim­ COLWELL, R. K. & J. A. CO DDINGTO N. 1994. Estimating terrestrial ilarities (Table 5) seemed strongly affected by distance biodiversity through extrapolation. Philos. Trans. R. Soc. Lon­ and habitat. Pakitza and Tambopata, both in south­ don 34.5:101-118. ERWIN, T L. 1982. Tropical forests: their richness in Coleoptera eastern Peru clustered, as did Onkone Gare, Ecuador and other species Coleopts. Bull. 36:74-7.5. and Neblina, Venezuela which are similar in habitat ---. 1983. Beetles and other insects of tropical forest canopies despite being approximately 1160 km distant from at Manaus, Brazil, samples by insecticidal fogging, pp ..59 - 7,5. In S. L. Sutton, T C. Whitmore & A. C. Chadwick (eds.), Trop­ each other. Taken together these western Amazonia ical rain forest ecology and management. Oxford, Blackwell Sci­ sites appear to form a region (Fig. ,3), perhaps because entific Publications. 498 pp. they lie along the eastern edge of the Andes. Porto FISIIER, E. M. 198.5 . A preliminary list of the robber flies (Diptera: Asilidae) of the Tambopata Reserved Zone, Madre de Dios, Velho was less faunistically similar to Sao Paulo de Peru . Rev. Peru. Entomol. 27:25-36. Olivenc;a than to Pakitza. HODKINSON, 1. D. & D. CASSON. 1991. A lesser predilection for These studies from Costa Rica to Bolivia demon­ bugs; Hemiptera (Insecta) dive rsity in tropical rain forests. Bioi. J. Linn. Soc. 43:101-109. strate that Site-specific data analysis is a prerequisite to LOUTON, J .A., R. W. GARRISON & O. S. F LINT. 1996. The Odonata a thorough understanding of regional biodiversity pat­ of Parque Nacional Manu, Madre de Dios, Peru; natural his­ terns. The methods presented above were useful in as­ tory, species richness and comparisons with other Peruvian sites, pp. 431-449. In D. E. Wilson & A. Sandoval (cds. ), La sessing biodiversity on a site by site basis, and once biodiversidad del sures te del Peru: Manu (Biodiversity of similar data from other studies and other organisms Southeastern Peru). Editorial Horizonte, Lima, Peru. 679 pp. are pooled, it may be possible to predict complemen­ MAY, R. M 1992. How many species inhabit the Earth? Sci. Am. 267(4):42-48. tarity between and among sites and to predict species PAULSON, D. R. 198.5. Odonata of the Tambopata Reserved Zone, numbers at other sites. Complementarity, or distinct­ Madre de Dios, Peru. Rev. Peru. Entomol. 27:9-14. PEARSON, D. J.. 198.5. The tiger beetles (Coleoptera: Cicindelidae) ness of species assemblages among sites can be used of the Tambopata Reserved Zone, Madre de Dios, Pel'll. Rev. with cluster analysis to predict complementarity be­ Peru. Entomol. 27:1.5-24. tween a wide variety of parameters such as biogeo­ POGUE, M. G. 1996. Biodiversity of Cicadoidea (Homoptera) of Pakitza, Manu Reserved Zone and Tambopata Reselved Zone, graphic, habitat differences, or host plant specificity. Peru: a faunal comparison, pp. 313-32.5. In D. E. Wilson & A. U sing species richness estimates, complementarity val­ Sandoval (eds.), La biodivcrsidad del sureste del Peru: Manu ues, and species lists generated from biodiversity in­ (Biodiversity of Southeastern Peru). Editorial Horizonte, Lima, ventories can be useful for biologists and conservation­ Peru. 679 pp. RAVEN, P. H. & E. O. WILSON. 1992. A fifty-year plan for biodiver­ ists to make more informed decisions about land use sity surveys. Science 258:1099-1100. and conservation. REAKA- KuDLA, M. L., D. E. WI LSON & E. O. WILSO N. 1997. Biodi­ versity II: understanding and protecting our biological re­ ACKNOWLEDGMENTS sources. The National Academy of Sciences. John HelllY Press, Washington, D.C . .5.51 pp. I would like to thank Teny L. ElWin for his support of this proj­ ROBBINS, R. K., G . LA MAS, O. H. H. MIELKE, D. J. HARVEY & M. M. ect and for providing timc in the field for coll ecting and processing CASAGR ANDE. 1996. Taxonomic composition and ecological material. For critically reviewing a draft of this paper, I thank Marc structure of the speCies-rich butterfly community at Pakitza, E. Epstein and Jerry A. LOll tan, Smithsonian Institution, Washing­ Parque Nacional del Manu, Peru, pp. 217- 252. In D. E. Wilson ton, D.C., and M. Alma Solis, Stuart H. McKamey and David R. & A. Sandoval (eds. ), La biodiversidad del sureste del Peru: Smith, Systematic Entomology Laboratory, Washington, D.C., Eric Manu (Biodiversity of Southeastern Peru). Editorial Horizonte, H. Metzl e r, Columbus, Ohio, and Deant' Bowers, University of Col­ Lima, Peru. 679 pp. orado, Boulder, Colorado. Funding was provided by the Smi th­ STORK, N. E. 1988. Insect diverSity: facts, fiction and speculation. sonian Institution's BTOLAT Program. BioI. J. Linn. Soc. 3.5:321-:337. SYSTAT. 1992. Statistics, version 5.2 edition. SYSTAT Inc., LITE RATURE CITED Evanston, Illinois. 724 pp. WILKE RSON, R. C. & G. B. FAIRCHILD. 198.5. A checklist and BU RN HAM, K. P. & W. S. OVERTON. 1978. Estimation of the size of generic key to the Tabanidae (Diptera) of Peru with special ref­ a closed population when capture probabilities vary among ani­ erence to tlle Tambopata Reserved Zone, Madre de Dios, Peru. mals. Biometlika 6.5:623- 633. K.ev. Peru. Entornol. 27:37-.53. - - - . 1979. Robust estimation of population size when capture probabilities VaIY among . Ecology 60:927-936. Received for publication 24 July 1998; revised and accepted 9 April C HAO, A. 19134. Non-parametric estimation of the number of 1999. classes in a population. Scaml. J. Statistics ll:265-270.